Testimonials versus Informational Persuasive Messages: The Moderating Effect of Delivery Mode and Personal Involvement OF TEOHNOLOGY by JUN 2 0 2008 Julia Braverman LIBRARIES PhD Social Psychology Northeastern University (2005) Submitted to the Harvard- MIT Division of Health Science and Technology in Partial Fulfillment of the Requirements for the Degree of Master of Science in Biomedical Informatics Ari at the -8 Massachusetts Institute of Technology, Cambridge, Massachusetts June 2008 Signature of Author .................................................... .. ............. Health Science an/ Technology Division May 7, 2008 Certified by .................................... 1 Qking Zeng Associate Professor, Harvard Medical School Thesis Supervisor A.cc.e. . .p.--ted b y . . . . . . . . . . . . . . . . . . . . . . . . . . . . "'V -Marta L. Gray Chairman, Department Committee on Graduat, Students Edward Hood Taplin Professor of Medical Engineering and Electrical En ineering Testimonial Versus Informational Messages Abstract Health communications use factual information or/and personal testimonials to inform and influence individual decisions that enhance health. Increasingly, Web and other computer-based systems are being used to communicate with patients. This study aims to test the relative effectiveness of testimonials compared to simple informational health messages presented through different modalities, and to the recipients with different levels of involvement. Results of the three independent experiments demonstrate that testimonials are more persuasive when presented through the audio mode rather than when presented through the written mode. Also, the informational messages are more persuasive when perceived by individuals characterized by high rather than low involvement and high rather than low need-for-cognition. The results are explained in terms of the Elaboration Likelihood Model (ELM). The interactive effect of transportation (Green & Brock, 2004) and involvement on persuasion is further examined. The findings help in developing the more effective ways of computer-based health communication. The highest level of efficiency can be achieved if the appropriate media modality and message format are used for recipients with certain initial involvement or need-for-cognition. Testimonial Versus Informational Messages Medical or health informatics is a rapidly growing discipline that is on the intersection between computer science, information science and health studies. It deals with optimizing methods of using technology to facilitate acquisition, storage, retrieval, and use of information in health and biomedicine. Consumer health informatics is the branch of medical informatics that focuses specifically on the consumers' (mostly patients') needs for information. It studies the methods of making information accessible to the customers, of facilitating doctor-patient communication and increasing the effectiveness of health promotions. Developing of both short and long-term technology-based behavioral health intervention is an integral part of consumer health informatics. Behavioral interventions deal with promoting cognitions, motivation and behaviors that are associated with improving health. Computer-tailored health intervention is the effective and relatively low-cost method of providing consumers with individualized feedback and advice (Kroeze, Werkman, & Brug, 2006). About fifty per cent of all mortality causes in the United States are linked to social and behavioral factors such as smoking, diet, drug use and insufficient physical activity. Yet less than five percent of the money spent annually on U.S. health care is devoted to reducing the risks of these preventable conditions. Behavioral and social interventions offer great promise, however their potential has been relatively poorly used. Increasingly, Web and other computer-based systems are being used to communicate with patients, to provide health information and to promote healthy behaviors. The effectiveness of those systems has been confirmed by many studies (Ammerman, Lindquist, Hersey, 2001). However, it is often hard to say, why a certain Testimonial Versus Informational Messages intervention is more effective than another one. Typically, the effect of interventions is evaluated "globally", without analyzing the specific strategies that lead to the substantial behavioral changes (Ammerman, Lindquist, Hersey, 2001; Kahn, Ramsey, Brownson et al, 2002). The goal of this research is to provide a detailed micro-analysis of several information delivery methods. All behavioral interventions, including the computer-based are heavily relied on health communication theories about the effective and persuasive messaging and methods of information delivery. Health communication is "the study and application of strategically designed messages delivered through selected media to convey relevance health information to targeted audience" (Kreps & Viswanath, 2001, p. x). The typical goals of health communication are to motivate a person to reduce risk (Werch, Grenard et al., 2006), to encourage screening and control (Rimer & Glassman, 1999), to increase the treatment adherence (Lerman et al., 1992), and so on. Effective communications can motivate people to engage in behaviors that will improve their health and enhance their quality of life. Health communications are substantially based on delivering persuasive messages about health and safety behaviors. Various persuasiveness techniques are employed to convince message recipients to stop smoking, to practice safe sex, or to follow a particular diet. One of the commonly used methods of information delivery in health communication is testimonial assertions or exemplars. Testimonials may include a personal story, a description of an individual experience, or a personal opinion. In a typical example of testimonial, a main character tells a story of his or her personal Testimonial Versus Informational Messages successful experience and directly or indirectly encourages the audience to follow her example; e.g.: "I began managing my eating habits four years ago. To me, the most rewarding thing is that I was able to lose weight and keep it off since my first semester." An alternative health communication technique is presenting factual information such as expository reports of events, expressing professional opinions or providing statistical evidence (Reinard, 1988). Most persuasion researchers have focused almost exclusively on factual or informational persuasive messages, whereas many communication and health practitioners widely exploit testimonials and anecdotal evidence messages (Slater, 2002). The relative effectiveness of each type of communication (where effectiveness refers to persuasiveness, attitude change, and behavioral compliance) has recently been investigated in literature, but the issue is still unresolved (see Baesler & Bargoon, 1994 for a brief review). Some studies found testimonials to be more persuasive than factual and statistical information (Brosius & Bathelt, 1994); other studies found the opposite to be true (e.g., Dickson, 1982). Many factors, including vividness (Baesler & Bargoon, 1994), emotional absorption (also labeled "transportation" degree) (Green & Brock, 2000), and message-value congruency (Slater & Rouner, 1996), were found to influence the relative persuasiveness of testimonials and informational messages. Many studies that investigated the relative persuasiveness of testimonials and factual information focused mainly on the message content (Baesler & Bargoon, 1994; Brosius & Bathelt, 1994; Slater & Rouner, 1996; but see Block & Kleer, 1997 for the effect of self-efficacy). In this research I focus on the recipient part of communication as well. The primary goal of this study is to show that the testimonial and pure informational Testimonial Versus Informational Messages message may be more or less effective depending on recipient's involvement, and also the modality by which the message was delivered (i.e., audio versus written message). Personali nvolvement Personal involvement has long been recognized as a primary factor that influences persuasion. According to the Elaboration Likelihood Model (ELM) (Petty & Cacioppo, 1979, 1979b), a recipient can be influenced by message content through one of two routes of persuasion: central or peripheral. When the route of persuasion is central, a recipient is influenced primarily by rational factors and information relevant to the message content. For example, while reading a message about a new diet, a recipient may consider whether the diet is healthy, if the diet's effectiveness is sufficiently supported by statistical evidence, whether the arguments are strong and logical, and if the source of information is reliable and can be trusted. When the route of persuasion is peripheral, other factors may influence a recipient's opinion. For the same diet message a recipient may consider source attractiveness, strength of emotional appeals (e.g., "Wow! It worked great!"), how enjoyable the message is, and if reading the message makes her feel happier. Personal involvement is an important factor in determining the route of persuasion (Petty & Cacioppo, 1979a; Petty, Cacioppo, & Goldman, 1981). Although there is a general agreement that involvement determines the recipients' motivation and attention to the topic (Kruglanski, Chen et al., 2006), it can be and has been operationally defined differently in various studies. For example, the involvement was operationalized as "personal relevance" (Johnson & Eagly, 1989, Petty & Cacioppo, 1979b, Petty, Cacioppo, & Goldman, 1981; Schul & Knapp, 1984; Sorrentino, Bobocel et al., 1988) and was induced by a reward expectation (Braverman, 2005). Whatever Testimonial Versus Informational Messages operationalization was used, those in the high involvement condition paid more attention to argument quality and were less affected by heuristic thinking. For medical interventions, involvement may be more of an individual rather than a situational factor. The importance of tailoring the message to specific characteristics of a recipient has long been recognized (see Rimer & Kreuter, 2006 for a brief history and review). Medical and health interventions also target a very different audience in terms of personal involvement. The current research makes a basic distinction between individuals who believe it is important to change their behavior (diet or alcohol drinking) and those who do not consider their current state to be a problem. This distinction may be compared to a distinction made by the stage-of-change model (Prochaska, DiClemente, & Norcross, 1992) between the precontemplation and other stages of behavior change. The goal of this research is to understand how an individual's initial readiness to change their behavior (and therefore his or her involvement with a message topic) predicts the individual's response to a informational reports versus a testimonial persuasive message. Rational decision suggests that information based on aggregated reports, which are more systematic and released by a credible source (as factual information often is), should be more persuasive compared to information based on a single piece of evidence (like a testimonial) (Baesler & Burgoon, 1994). The validity of factual information can be regarded as high relative to the testimonials (Brosius & Bathelt, 1994). However, many researchers found that single-report evidence, a vivid exemplar, or a testimonial were more persuasive than statistical evidence (Brosius & Bathelt, 1994). This suggests that the persuasive effect of testimonials is based not on systematic, but rather on heuristic thinking. The theory of exemplification (Zillmann & Brosius, 2000) also states that the Testimonial Versus Informational Messages effect of testimonials (or exemplars) is based on cognitive shortcuts such as availability and representativeness heuristics. The theory maintains that testimonials are very efficient in public media because individuals typically do not engage in extensive cognitive processing of media messages. The following questions can then be asked: What if individuals do engage in extensive message processing? Will this affect the persuasive power of testimonials versus fact-based informational messages? As a factor that determines the likelihood of cognitive processing, personal involvement can be an important moderator that explains the comparative effectiveness of fact-based informational versus testimonial messages. Reliance on heuristics (as in testimonials) influences persuasion especially when it occurs through a peripheral route. On the contrary, taking factual information into account is more characteristic of the central route of persuasion. This might be compared to the effect of message strength on persuasion. In the paper of Petty, Cacioppo & Goldman (1981) the difference between strong and weak message was described as follows: "the strong version of the message provided persuasive evidence (statistics, data, etc.) In contrast, the weak version of the message relied more on quotations, personal opinion, and examples to support its position." (p. 850). According to the ELM, "strong" messages are more convincing when the central route of persuasion is activated. Whereas, when the peripheral route is activated no effect of message strength is usually found. One can thus hypothesize that when personal involvement is high and consequently the central route of persuasion is exploited, informational message will be more persuasive than testimonials. However, when personal involvement is low and a peripheral route of persuasion is used; the testimonial message will be more or equally persuasive compared to the informational message. Delivery modality Testimonial Versus Informational Messages Another factor that can moderate the effectiveness of testimonial versus fact- based informational message is the delivery modality. The majority of studies that investigated persuasion including those that compared testimonial stories and fact-based information (Green & Brock, 2002) dealt only with written forms of a message. However, the same message can be delivered through a multitude of modalities. For example, it can be presented as a text (written mode) or as a recording (audio mode). Bathelt (1994) compared the effect of radio versus textual message presentation on the perceived distribution of opinions. It was hypothesized that the effect of exemplar on judgments would be greater in radio presentations compared to print presentations. The author suggested this because radio exemplars can be more vivid than exemplars in print (ibid.). However, this hypothesis was only partially supported. Although indeed some researchers find no difference between message modes (McGinnies, 1965, Tannenbaum & Kerrick, 1954), others have reported that modality can affect persuasion both alone (Haugh, 1952; Knower, 1935) and in interaction with other variables (e.g., source credibility and argument quality) (Andreoli & Wrochel, 1978; Booth-Butterfield & Gutowski, 1994; Chaiken & Eagly, 1983; Worchel, Andreoli, & Eason, 1975). Studies consistently show that message modality affects message processing. Print or written modality makes systematic message processing more likely, whereas the effect of peripheral or heuristic cues is more pronounced when the message is delivered through audio or video mode (Andreoli & Worchel, 1978; Brooth-Butterfield & Gutowski, 1994). According to the ELM, those recipients who use a peripheral route of persuasion should be more affected by heuristic cues than the recipients who use a central route of Testimonial Versus Informational Messages persuasion. Hence, low-involved individuals are expected to be persuaded more by an audio message than by a written text. At the same time, highly involved individuals should be less affected by the modality. Moreover, one can expect that the testimonial form of the message will be more persuasive when recorded rather than just written, because the additional emotional information can be conveyed by a speaker's tone of voice potentially increasing transportability and vividness (Bathelt, 1994) of a testimonial story when spoken rather than written. Study 1 For the first study I composed a persuasive message about drinking water to lose weight. The topic was chosen because people vary significantly in the degree to which the topic of weight loss is relevant to them. Drinking water also seems like plausible and easy-to-follow advice. Pure informational and testimonial versions of the same message were prepared in both written and audio format. The formal (informational) text looked like a standard public-oriented health promotion communication (e.g. those available on WebMed website). The testimonial version presented a personal story of a student who tried drinking a large amount of water and lost weight. Participants reported to what degree weight loss was important to them, and to what degree they were persuaded by the message. The following hypotheses were stated: Hypothesis 1. Participantsw ho report being highly motivated to change their diet will be more persuaded by the informationalm essage, whereasp articipantsw ho report being low in motivation to change their diet will be more persuadedb y the testimonial.